Which is great - it just felt like there was some disagreement between whether it's actually a problem to be fixed :- "let's fix the local repository" vs "Yes, I'm just saying separate processes should be using separate local repositories." which I didn't understand as I couldn't see how a locally built artifact would then be able to be accessed by downstream builds.
Is there a JIRA covering this one (I guess there's probably several ;-) ) - MNG-3151 seems to me to be a slightly different thing from what we're now talking about. On 29/08/2007, Brett Porter <[EMAIL PROTECTED]> wrote: > > I already responded to this here: http://www.nabble.com/Re%3A-local- > repositories-%28was%3A-http%3A--jira.codehaus.org-browse-MNG-3151%29- > p12286344s177.html > > Cheers, > Brett > > On 29/08/2007, at 4:49 PM, Nigel Magnay wrote: > > >> > >> If you are knowingly doing this then I would say different settings > >> via a build plan to prevent corruption or the CI mechanism serializes > >> the builds. > >> > > > > How can one do this and still make use of all 8 cores to build on a > > desktop machine (or all 16 on our CI server)? m2 builds could run > > really parallel, but they frequently break when they do (particularly > > when one buildl downloads a plugin, and another one uses it half-way > > through, getting wrong metadata). > > > > Is there a way of serializing access to the local repository? > > > > This seems a really important feature given the way computers are > > rapidly adding cores. m2 has the intelligence to know what things can > > be parallelized because of its smart dependency mechanism, but it > > falls down because of the dumb local repo into an antlike serial > > process. > > > > --------------------------------------------------------------------- > > To unsubscribe, e-mail: [EMAIL PROTECTED] > > For additional commands, e-mail: [EMAIL PROTECTED] > > --------------------------------------------------------------------- > To unsubscribe, e-mail: [EMAIL PROTECTED] > For additional commands, e-mail: [EMAIL PROTECTED] > >
